Windmolens en warmtepompen
Transcription
Windmolens en warmtepompen
Distributed Control Concepts using Multi-Agent technology and Automatic Markets An indispensable feature of smart power grids Maarten Hommelberg, Cor Warmer, Rene Kamphuis, Koen Kok, Gerrit Jan Schaeffer www.ecn.nl Why decentralized control? • Privacy • Autonomy • Scalability • Communication overhead Three stages of DG Growth Growing DG Penetration Accommodation • DG accommodated • • • in the current system DG units running free DG treated as negative demand Central control unchanged Decentralization • Added value of • • • clustered control of DG. Common ICT systems: Virtual Utilities, Virtual Power Plants. Central control still needed Decentralized, bottom-up control is added. Dispersal • Distributed power • • • dominates the market Network of networks Local network segments selfsupplying. Central controller becomes a coordinator. • Source: IEA, 2002 Current practice in load control • Centralized control • Reactive customers • Traditionally focused on customer demand • Undetermined outcome PowerMatcher control • Decentralized control • Active customers • Transparency with respect to demand and supply • Determined outcome by ‘real-time’ contract • Dynamic response PowerMatcher cells - networks of networks Business cases • • • • • • • Imbalance reduction for responsible party (CRISP) Keeping self-generated power in-house (UPM) Flattening generation and consumption patterns Intelligent substation control: peak reduction (SPS First trial); towards islanding Virtual Power Plant control (FENIX) Storage optimization (plug-in hybrids) … CRISP Field test: Portfolio 2.5 MW 15 kW Local CRISP-Node Central CRISP-Node Wind Turbine Park I Cold Store Scaled up by simulation to 1,5 MW Local CRISP-Node 8 MW Local CRISP-Node Data Communications Network Local CRISP-Node Wind Turbine Park II Local CRISP-Node 6 MW Residential Heat Production (CHP) 50 kW Emergency Generator Local CRISP-Node 0.8 kW ECN Test Dwelling Scaled up by simulation to 800 kW CRISP Æ 43.5 % imbalance reduction SPS First trial ... Local VPP-Node Local VPP-Node Local VPP-Node GPRS Wireless Communication Local VPP-Node Central VPPController GPRS Supplier: SPS First trial Winter Situation comparison space temperature bandwidth 14000 SubstationLoad_Substation Conv Substation Load Substation Load SDM 0.5 Substation Load SDM 1 Substation Load SDM 2 SDM Substation load 5 12000 10000 8000 Load [W] 6000 4000 2000 0 1 101 201 301 401 -2000 -4000 -6000 Time ste ps [15 minute s] 501 601 FENIX (Northern demo) Overall System Architecture DEMS DEMS Application SCADA Server Web Access VPP Concentrator Box 4 W ind generation I n te r n e t Woking Council & EDF Energie Nouvelle EDF Energy G1 Woking Swimming Pool & Imperial College Labs G1 G1 G1 Fenland Glass Moor Windpark (8 Re-Power MM2 2MW each – total 16MW) VPP Concentrator 1 & 2 VPP Concentrator 3 (Existing CHP SCADA) Fenix Box Load under Demand Response L2 Distributed Balancing Cluster G1 G3 PV Cell 2 Load under Demand Side Management > Fen ix N or th er n Scen a ri o – A REVA Pr op osa l Horsell SS L1 Woking SS 33KV 11KV Old Woking SS G2 Woking CHP Portfolio (total 4MW) 2 UPM • Connectivity for: • Small scale wind • PV(T) • µCHP • Heat pumps • Household appliances • UPS Functionality on a household level • Connectivity with other UPMs Upcoming research • Integral • Active houses • 6 scenario simulations • Network constraint handling Conclusions • Distributed control concepts create: - scalability - local autonomy - market integration • The PowerMatcher provides a flexible concept to implement a variety of business cases • Different business cases proven in Field tests • Embedded trajectory as a first step to commercial products